| |
Autumn 2011 IERG5154 Information Theory
Page history
last edited
by sidjaggi 13 years ago
Welcome to IERG5154 Information Theory!
Final Paper
Here are some official documents provided by the University. Please read them in advance.
Student/Faculty Expectations on Teaching and Learning
http://www.erg.cuhk.edu.hk/erg-intra/upload/documents/StaffStudentExpectations.pdf
Academic honesty and plagiarism
|
Attention is drawn to University policy and regulations on honesty in academic work, and to the disciplinary guidelines and procedures applicable to breaches of such policy and regulations. Details may be found at http://www.cuhk.edu.hk/policy/academichonesty/ . With each assignment, students will be required to submit a statement that they are aware of these policies, regulations, guidelines and procedures.
|
Course Information
Instructor: Professor Sidharth (Sid) JAGGI
jaggi@ie.cuhk.edu.hk
http://staff.ie.cuhk.edu.hk/~sjaggi/
Calendar: http://calendar.jaggi.name
TA: Fan CHENG,
cf008@ie.cuhk.edu.hk
Office Hour: Monday 2:00-3:00, SHB729
Time and Location: T 2:30-5:15pm SHB 833
Learning resources for students
Related courses
1. 1st Sem: IERG6120 (Network Information Theory)
2. 2nd Sem: IERG6130 (Network coding)
|
Course Title: IEG 5154 Information Theory
|
|
Description: The course aims to cover
- Fundamental definitions of information measure (entropy, conditional entropy, mutual information) and their properties.
- Lossless Source Coding/Data Compression -- theory and algorithms
- Channel Coding/Error-Correcting Codes/Coding theory
- Rate-distortion theory.
-
Miscellaneous advanced topics depending on time and interest (Kolmogorov Complexity, Universal Portfolio Theory, Network Coding, ...)
|
Content, highlighting fundamental concepts (not necessarily in the chronological order we will use in classroom discussions)
|
Topic
|
Contents/fundamental concepts
|
- Information theoretic quantities
|
- Entropy, conditional entropy, mutual information, divergence, differential entropy, Markov/ergodic sources, properties (chain rules, positivity, convexity, Jensen's inequality, Fano's inequality, conditioning, data processing inequality, Law of large numbers, Sanov's theorem, AEP, entropy rates)
- Achievability and converse proofs, Kraft inequality, codes -- Shannon-Fano-Elias, Huffman, Arithmetic, Lempel-Ziv (universal codes)
- Hamming codes, channel capacity (achievability and converse proofs), zero-error capacity, joint source-channel coding, feedback capacity, Gaussian channels (parallel channels, coloured noise)
- Scalar/vector quantization, rate-distortion theorem (achievability and converse)
|
|
Learning outcomes:
- Demonstrate ability to manipulate basic information-theoretic quantities to prove relevant theorems.
- Demonstrate understanding of foundational topics in information theory, and an ability to use the theoretic tools required to prove corresponding theorems.
- Use the above to characterize and design information storage, manipulation and transmission systems.
|
Learning activities
|
Lecture
|
Problem Sets
|
Online Activities (Scribe Notes/Discussion)
|
Homeworks
|
|
(hr)
in class
|
(hr)
in/out class
|
(hr)
out of class
|
(hr)
out of class
|
|
36
|
0
|
36
|
12
|
12
|
6
|
15
|
0
|
|
M
|
O
|
M
|
O
|
M
|
O
|
M
|
O
|
M: Mandatory activity in the course
O: Optional activity
NA: Not applicable
Assessment scheme
|
Task nature
|
Description
|
Weight
|
|
Problem sets
Homework
Scribe Notes
Class participation
Final Exam
|
In-class problems, handed in next class
Collaborative homework
Scribe notes of a particular lecture
In-class discussion/Discussion on wiki
Examination
|
20% (8)
20% (4)
20% (2)
15%
25%
|
Homework Policy: Answer all the questions according to the lecture progress and hand in them in the next class.
The weekly problem set is cancelled.
Feedback for evaluation:
Class evaluations
Students are welcome to express their comments and suggestions via the following formal and informal feedback channels:
- Two course evaluations. First one to be conducted in the middle of the term and the second one at the end of the term. Students are encouraged to provide specific comments and/or suggestions in addition to the numeric ratings.
- At the end of each lecture there will be a single question feedback slip given to each student.
- Students are also encouraged to provide feedbacks using informal channels, such as email/discussion to instructor/tutor, and via the talk pages on the class wiki.
Tentative Course Schedule (will edit as we go along)
| DATE |
TOPIC |
READINGS |
|
Logistics/Introduction. Basic probability theory. Worst-case compression. Binary trees. |
|
|
Method of types, typical/atypical sets, sizes/probabilities, Sanov's Theorem |
11.1-11.5, Cover/Thomas |
|
Sanov's Theorem -- Achievability/Converse of Source Coding Theorem
|
11.1-11.5, 2.1-2.6, 2.8, 2.10, Cover/Thomas |
|
Entropy definitions in terms of typical sets, properties. |
11.1-11.5, 2.1-2.6, 2.8, 2.10, Cover/Thomas |
|
Properties of entropy-type functions. |
2.1-2.6, 2.8, 2.10, Cover/Thomas |
| |
Information-theoretic proof of Source Coding Theorem |
7.1-7.7, 7.9, Cover/Thomas |
|
Discussion of Channel Coding Theorem |
7.1-7.7, 7.9 Cover/Thomas |
|
Proof of Channel Coding Theorem |
7.6-7.7,7.9 Cover/Thomas |
|
Classifications and Properties of Per-Symbol Source Coding Schemes |
5.1-5.5 Cover/Thomas |
| |
Shannon Code, Huffman Code |
5.6 Cover/Thomas |
|
Entropy Rates of a Stochastic Process |
4 Cover/Thomas |
|
Arithmetic Codes, Weak Typicality |
5.9, 13.3, 3 Cover/Thomas |
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
| |
|
|
Lectures/Problem Sets/Homeworks/...
Important Announcements:
Item
|
Date
|
Topic
|
Author
|
Details
|
Homework 1
|
2011/09/30 |
|
|
We update the description of problem 1 in homework 1
|
Lecture on 10/4
|
2011/10/03
|
|
|
We will have lecture on 2011/10/04
|
Final
|
2011/11/22
|
|
|
The take-home final will start at 11:59 a.m. Friday, 2011/12/16. The duration is 24 hours.
|
Course Review
|
2011/11/22
|
|
|
We will have a class at 2 p.m.-5 p.m. Thursday, 2012/12/08. Room833
|
Final
|
2011/12/12
|
|
cheng fan
|
The final will last for 72 hours from Friday noon (11:59 a.m.) to Monday noon (11:59 a.m.).
It is open notes/book -- cover/thomas, closed internet/collaboration.
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
Scribe Notes Schedule:
| Week |
Tuesday |
|
| 1 |
|
|
| 2 |
Siyao, WANG Limin
|
|
| 3 |
Yang liu , Zirui Zhou, Chi Zhang, WONG Pak Kan
|
|
| 4 |
LEONG Ho Ka, CHAN Chun Lam, FONG Cheuk Man Manson
|
|
| 5 |
YE Jihang, FENG Yiyong
|
|
| 6 |
Yuen Piu Hung, Wang Qike, LIU Zhongchang
|
|
| 7 |
Yichao Li, Jiang Yunxiang
|
|
| 8 |
Yip Kit Sang Danny, Lee Chin Ho,
|
|
| 9 |
TANG Wanrong, YAO Leiyi, Hu Xihao
|
|
| 10 |
CAI Sheng, TU Jinlong ,Zhan Lei |
|
| 11 |
LU Lian, BI Suzhi Wu Xixuan, LU Tan
|
|
| 12 |
Luk Hon tung
|
|
Free help:
1. Learn how to use PBwiki: The PBwiki Manual
2. If you prefer video, watch a recording of our popular webinar, PBwiki 101: Your Guide to Wiki Basics.
3. Need more help? Sign up for a Free introductory webinar
4. Not So Short Introduction to LaTeX
Autumn 2011 IERG5154 Information Theory
|
|
Tip: To turn text into a link, highlight the text, then click on a page or file from the list above.
|
|
|
|
|
Comments (14)
sidjaggi said
at 6:01 pm on Oct 4, 2011
Two comments/classes of comments stood out today.
One pointed out that the workload is perhaps unreasonable -- in particular, requiring every problem set to be turned in (in addition to scribe notes/homeworks/exam) is a lot.
henceforth, you don't need to turn in problem sets -- the grades for problem sets will be redistributed to the homeworks.
the second class of comments, which is a reprise of comments i've seen before as well, is that many of you would like me to also formally write down definitions/statements/theorems/etc.
valid point. my own style tends to be to wax eloquent about the intuition verbally, and get carried away, and forget to write things down. this can be a disadvantage when it comes to abstract concepts. i'll try to keep this in mind, but whenever i forget, i'm going to request that those of you who find this problem, please raise your hand and gently request that i write down the appropriate definition/formula/whatever...
others liked the review/intuition/etc. i liked that you liked it :)
thanks for the feedback, and please keep it coming!
sidjaggi said
at 10:30 pm on Oct 5, 2011
On further thought, instead of jus cancelling the problem sets, we'll substitute it with a two-hour in-class mid-term on the 25th of October.
Shiney Wanrong TANG said
at 5:03 pm on Oct 10, 2011
Dear prof,
Actually, most of us would prefer to hand in problem sets rather than have a mid-term exam. In addition, we have already handed in two or three sets.
Regards,
Wanrong
ZhanLei_Jacky said
at 5:08 pm on Oct 10, 2011
Hmm, I think so, too~~
ZhanLei_Jacky said
at 6:00 pm on Oct 10, 2011
I mean I prefer problem sets~~
Jiang Yunxiang said
at 6:54 pm on Oct 10, 2011
I agree
sidjaggi said
at 1:46 am on Oct 11, 2011
hmm, ok, let's do it this way. Those who prefer problem sets can continue to do problem sets instead of the midterm. Those who prefer the mid-term can do the midterm in lieu of the problem set. You have to (in-class) tell the TA which you'd prefer.
The midterm will nonetheless occur on the 25th (since I'm traveling that week anyway, and will not be able to teach then -- the TA will do a 45-minute in-class review session, followed by a 2 hour midterm for those who want it).
sidjaggi said
at 1:48 am on Oct 11, 2011
Also, even if you choose the midterm, but like turning in problem sets since they give you more practice, we'd be happy to look at them and grade them for you (without those grades changing your score).
Chin Ho Lee said
at 5:19 pm on Oct 10, 2011
I'd prefer midterm...
sidjaggi said
at 2:36 pm on Oct 7, 2011
A mistake, and a glaring omission, in my discussion on Tuesday. Anyone who emails me with both before the next class gets an automatic grade bump... ;)
Yip Kit Sang Danny said
at 12:04 pm on Dec 16, 2011
Where is the final exam paper?
sidjaggi said
at 2:03 pm on Dec 16, 2011
Just updated Q1 of final to make part a easier (1400 hours Friday)
Xihao said
at 3:17 pm on Dec 16, 2011
For question 1:
Where does the F_2 come from?
Whether c is positive or negative in formula 1-2^{c|m-n|}?
Three sub-questions have point 2, point 3 and point 3, but their sum is 7 points!
sidjaggi said
at 5:02 pm on Dec 16, 2011
Thanks for pointing those out. Updated question paper with corresponding (minor) changes. In particular:
1. F_2 is the binary field. As the question says, your are expected to use a binary linear code.
2. c is positive.
3. Good point(s :) Corrected. Total points now 8.
You don't have permission to comment on this page.